Biogeography of the xerophytic genus Anabasis L. (Chenopodiaceae)

Abstract Aim Using the extremophile genus Anabasis, which includes c. 28 succulent, xerophytic C4 species, and is widely distributed in arid regions of Northern Africa, Arabia, and Asia, we investigate biogeographical relationships between the Irano‐Turanian floristic region (ITfr) and its neighboring regions. We test whether the spread of arid and semi‐arid biomes in Eurasia coincides with the biogeography of this drought‐adapted genus, and whether the ITfr acted as source area of floristic elements for adjacent regions. Location Deserts and semi‐deserts of Northern Africa, Mediterranean, Arabia, West and Central Asia. Methods Four cpDNA markers (rpL16 intron, atpB‐rbcL, trnQ‐rps16, and ndhF‐rpL32 spacers) were sequenced for 58 accessions representing 21 Anabasis species. Phylogenetic relationships and divergence times were inferred using maximum likelihood and a time‐calibrated Bayesian approach. To document the extant distribution of Anabasis, material from 23 herbaria was surveyed resulting in 441 well‐documented collections used for the coding of eight floristic regions. Using this coded data, ancestral range was estimated using “BioGeoBEARS” under the DEC model. Results Anabasis originated during the Late Miocene and the ancestral range was probably widespread and disjunct between Western Mediterranean and the Irano‐Turanian regions. Diversification started with two divergence events at the Miocene/Pliocene boundary (5.1 and 4.5 mya) leading to Asian clade I with ITfr origin which is sister to a slightly younger Asian clade II, which originated in the Western ITfr, and a Mediterranean/North African clade with an origin in the Western Mediterranean. Main conclusions Anabasis did not follow aridification and continuously expanded its distribution area, in fact its probably wide ancestral distribution area seems to have been fragmented during the very Late Miocene and the remnant lineages then expanded into neighboring arid regions. This genus supports the role of the ITfr as source area for xerophytic elements in the Mediterranean and Central Asia.


| INTRODUC TI ON
The Irano-Turanian floristic region (ITfr) as defined by Griesebach (1884) and Takhtajan (1986) covers c. 30% of Eurasia and ranges from southern parts of Mongolia and western provinces of China, Kyrgyzstan, Tajikistan, Pakistan, Afghanistan, southern parts of European Russia, Kazakhstan, Uzbekistan, Turkmenistan, Iran, and Iraq to the Anatolian plateau, inland parts of Syria and Lebanon, and Jordan. The ITfr harbors more than 27,000 species in its species-rich western part and around 5,000 species in its eastern part (Manafzadeh, Staedler, & Conti, 2017 and ref. therein). The degree of endemism in the ITfr ranges between 20%-40% (Takhtajan, 1986;Zohary, 1981) and is particularly high in the three biodiversity hotspots of the western ITfr: the Irano-Anatolian region, the Mountains of Central Asia, and the Caucasus (see Manafzadeh et al., 2017;Solomon, Shulkina, & Schatz, 2013). Among a number of features described as characteristic for the ITfr is the high diversity of Chenopodiaceae (sensu Walker et al., 2018), especially in desert and semi-desert areas (summarized in Djamali, Brewer, Breckle, & Jackson, 2012;Manafzadeh et al., 2017). In these arid areas, the vegetation is dominated by a high number of C 4 chenopods species (Manafzadeh et al., 2017;Schüssler et al., 2017;Takhtajan, 1986). C 4 photosynthesis is a recently evolved elaboration of the conventional photosynthetic carbon reduction cycle, also known as C 3 pathway, to concentrate CO 2 for utilization by ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) in the Calvin cycle (Hatch, 1987).
Aridification in the ITfr started during the Eocene-Oligocene transition and intensified during the Middle Miocene-Pliocene (Zhang et al., 2014). In this latter phase, uplifts of mountain chains and plateaus (e.g., Alborz, Tien Shan, Zagros) caused large rain shadows, continuous temperature decrease, and increased continentality, which likely triggered the expansion of xerophytic plant communities in the ITfr (Manafzadeh et al., 2017 and ref. therein). According to Djamali et al. (2012), the three climatic factors, continentality, winter temperature, and precipitation seasonality, differentiate the ITfr from its adjacent territories, the Mediterranean, the Saharo-Arabian, Euro-Siberian and the Central Asiatic regions. Among these three factors, continentality was found to be the prime factor that separates the ITfr from Mediterranean and Saharo-Arabian regions and also the main factor separating sub-regions within the ITfr itself (Djamali et al., 2012).
Based on floristic similarities, a close relationship of the ITfr to the Mediterranean region and Saharo-Arabian region has long been proposed (Takhtajan, 1986;Zohary, 1973). Consequently, some authors hypothesized that the ITfr served as a source area for the adjacent floristic regions (Comes, 2004;Djamali et al., 2012;Manafzadeh, Salvo, & Conti, 2014;Manafzadeh et al., 2017;Roquet et al., 2009;Zhang et al., 2014;Zohary, 1973), mostly because a stable dry climate has persisted in some parts of the ITfr since the early Eocene, hence providing a stable habitat for plant lineages over a long time (Manafzadeh et al., 2014(Manafzadeh et al., , 2017. Studies in Apiaceae (Banasiak et al., 2013), Brassicaceae (Franzke, Lysak, Al-Shehbaz, Koch, & Mummenhoff, 2011;Karl & Koch, 2013), and Rutaceae (Manafzadeh et al., 2014) support this hypothesis. However, only few molecular, historical biogeographic studies have so far been conducted that rigorously tested relationships between the ITfr and recipient areas as well as possible dispersal events or migration routes. In particular, the biogeographical study of the xerophytic Haplophyllum A. Juss.
(Rutaceae) supported the role of the western ITfr as a source area for xerophytic elements found in the Mediterranean (Manafzadeh et al., 2014). Though, additional studies of the ITfr plant lineages are needed to test a putatively source-like character of the ITfr using biogeographical analyses of dated phylogenies in order to put divergence and diversification into time and space.
As a monophyletic lineage within the ITfr typical element Salsoleae-Chenopodiaceae, with a proposed stem age dating back to the Miocene (Schüssler et al., 2017), the xerophytic genus Anabasis L. is suitable to investigate the relationships of xerophytic elements of the ITfr and its adjacent regions. According to literature and flora treatments, Anabasis is widely distributed in steppes, semi-deserts and deserts of North Africa, West and Central Asia (Hedge, 1997;Sukhorukov, 2008), and it also occurs in the most southern parts of Spain, the Eastern Mediterranean, South Siberia, West China, and Mongolia. Hence, with this wide distribution Anabasis covers not only the entire ITfr but is also present in most adjacent floristic regions, thus a perfect candidate genus to infer the floristic relationships among these areas and eventually to test whether the ITfr acts as source area for adjacent regions.
However, the current assessment of the distribution of Anabasis species is relatively rough and likely incomplete.
Here, we conducted a survey of c. 600 available herbarium specimens of 28 species of Anabasis to infer their distribution areas. Using a resolved and dated molecular phylogeny based on 58 accessions representing 21 species of Anabasis and data from four chloroplast markers, its biogeographic origin and expansion in the ITfr adjacent regions were reconstructed to test whether the ITfr served as a source of species to the recipient regions, and whether Anabasis followed the spread of arid biomes in Eurasia and North Africa.   (Miller, Pfeiffer, & Schwartz, 2010). Based on the Akaike information criterion (AIC), the best fitting model was the GTR+γ model. The ML analyses were carried out using RAxML v.8 (Stamatakis, 2014).

| Phylogenetic inference and molecular dating
Calibration of the molecular clock and calculation of divergence times were performed using BEAST v.2.4.5 (Bouckaert et al., 2014) on CIPRES Science Gateway v.3.3 (Miller et al., 2010). The BEAST xml input files were created with BEAUti v.2.4.5 (Bouckaert et al., 2014). Outgroup (Suaedoideae and Salicornioideae) as well as the ingroup (all others) was treated as monophyletic and the age of the most recent common ancestor (tmrca) for the ingroup was calibrated using a normal distribution prior with a mean of 30.75 and sigma of 5.55, matching the 95% highest posterior density (HPD; 39.9-21.6 mya) of Kadereit, Newton, and Vandelook (2017).
For the BEAST analysis, we used the substitution model GTR+γ with four gamma categories. The uncorrelated lognormal relaxed clock under a Birth-Death speciation process (Gernhard, 2008;Nee, May, & Harvey, 1994) with a random starting tree was set for the molecular dating analysis. The Markov chain Monte Carlo (MCMC) ran for 50 million generations and sampling every 5,000 generations. The performance of the BEAST run was checked in TRACER v.1.6 (Rambaut, Suchard, & Drummond, 2014) using the BEAST log file. The first 10 percent of the sampled trees were discarded as "burn-in." The remaining trees were summarized using TREEANNOTATOR v.2.4.5 (Bouckaert et al., 2014), and 95% confidence limits for ages of the nodes were calculated.

| Biogeographic analyses and species distribution
The assessment of the distribution of the species was based on a survey of c. 600 herbarium specimens which were loaned from B,

Samples in molecular phylogenetic analysis (corresponding to Chen No. in Supporting Information Appendix S1); samples used in the biogeographical analysis in bold
No of specimens included in the assessment of distribution area (Supporting Information Appendix S2)  (Takhtajan, 1986) Table 1; Figure 2). The ITfr is represented by the regions D, E, F, G, and southernmost part of H.

Coding for biogeographical analyses
For the biogeographical analyses, another BEAST analysis was performed using nearly the same settings as above but with a reduced data set that included only one accession per species to avoid any errors due to sampling bias, that is multiple accessions of some species versus only one accession in other species. For monophyletic species, the accession with the most sequence information available was included in the analysis, while for the four polyphyletic species two accessions per species were used for the analysis (see Table 1; Results section). The calibration derived from the first BEAST analysis for the crown node of Anabasis (excl. A. ehrenbergii) was used (normal prior with mean of 5.21 and sigma of 1.79, 95% HPD: 8.14-2.26 mya), and a MCMC of 25 million generations sampling every 2500 generations. Ancestral range estimation (ARE) was conducted using "BioGeoBEARS" (Matzke, 2013(Matzke, , 2014
The ML analysis (not shown) and the Bayesian analysis (see Supporting Information Appendix S4) revealed identical topologies. Anabasis (excl. A. ehrenbergii Schweinf. ex Boiss.) is monophyletic with high support.
Anabasis ehrenbergii is solved as sister to the remaining Anabasis species, albeit with only low support (posterior probability 0.94). In previous studies with less accessions of Anabasis, A. ehrenbergii was in an unresolved position among other members of the Salsoleae (Schüssler et al., 2017). The position of Anabasis within Salsoleae still remains poorly resolved (as in Schüssler et al., 2017). For all species except A. cretacea Pall., multiple accessions from different regions were included and all but four species are resolved as monophyletic (Figure 3).
The four species that are probably not monophyletic are A. aphylla L.,  Figure 3, Supporting Information Appendix S4).

| Biogeographical analyses
Based on the likelihood and AIC values, the best fit model was the DEC model (Table 2)

| D ISCUSS I ON
The ITfr has been suggested to be the geographical origin of, for example, the family Brassicaceae (Franzke et al., 2011;Karl & Koch, 2013) or tribe Cardueae, Compositae (Barres et al., 2013). Also, Jabbour and Renner (2011) could show strong biogeographical links between the ITfr and the Mediterranean region in tribe Delphinieae (Ranunculaceae). Furthermore, even if not the geographical origin, the ITfr was proposed to be a major center of diversification in subfamily Apioideae, Apiaceae (Banasiak et al., 2013) or the Campanula alliance, Campanulaceae (Roquet et al., 2009). Besides these examples of plant groups inhabiting rather temperate habitats, the ITfr was suspected as the likely source area especially for arid taxa found in neighboring regions, in particular in the Mediterranean area (Blondel, Aronson, Bodiou, & Boeuf, 2010;Comes, 2004;Quézel, 1985;Takhtajan, 1986;Zohary, 1973). Arid regions play an essential role for terrestrial biomes, as the desert and semi-desert biomes occupy together more than one-third of the global land surface (Laity, 2008 (Manafzadeh et al., 2014(Manafzadeh et al., , 2017  from North Africa to Central Asia, is highly adapted to aridity, and so is an excellent model taxon to further infer the biogeographic relationships of xerophytic elements of the ITfr and its adjacent regions. Georeferencing of 441 herbarium specimens of Anabasis showed that the distribution area of the genus covers large parts of these arid areas (Figures 2,5). The relatively low total number of Anabasis collections with sufficiently documented localities was compiled by an exhaustive investigation of the material of 23 herbaria. This clearly indicates that most of these desert areas are poorly represented in herbarium collections and might partially explain why xerophytes of the ITfr have been poorly studied. Fifteen of the 28 spp. studied (Table 1)  While most species are restricted to one or two floristic provinces, only two species, A. setifera (Figure 5a: violet squares) and A. syriaca is questioned by tree topologies resulting from nuclear data sets (Schüssler et al., 2017). For seven species (Table 1) Interestingly, the biogeography of Haplophyllum (Manafzadeh et al., 2014) shows parallels to Anabasis: Both Haplophyllum and Anabasis started diversifying at the very end or shortly after the Messinian salinity crisis at the end of the Miocene (Rouchy & Caruso, 2006). Also, during the end of the Miocene, Asian Zygophyllum (Zygophyllaceae), which is another arid-adapted element of Central Asia, underwent a burst of diversification (Wu et al., 2015). This is a remarkable result, because in contrast to Haplophyllum and Asian Zygophyllum, which likely originated in the Early Eocene and Early Oligocene, respectively (Manafzadeh et al., 2014;Wu et al., 2015), elements of Central Asia, tolerating extreme drought (Kürschner, 2004 photosynthesis (Schüssler et al., 2017;pers. observation). Several species are able to resprout (e.g., Bokhari & Wendelbo, 1978;Fahn & Dembo, 1964;Olufsen, 1912;Sukhorukov & Baikov, 2009;Voznesenskaya, 1976a,b;pers. observation). Studies of the reproductive organs of Chenopodiaceae show that Anabasis seeds have large, green, coiled embryos without nutritive tissue that is in agreement with the seed structure of other Salsoloideae (Sukhorukov, 2008;Sukhorukov et al., 2015) having very fast germination (Kadereit et al., 2017 and ref. therein). Climate change was shown to differently affect regions of the world (Kirtman et al., 2013). For the ITfr, it was projected that the effects will vary depending on the location within the ITfr: precipitation will increase in some parts of the ITfr, whereas it will decrease in other parts (Kirtman et al., 2013;Manafzadeh et al., 2017). The slow-growing Anabasis is highly specialized in arid habitats and likely is at a competitive disadvantage under more mesic conditions (see above). Thus, arid-adapted lineages of the highly diverse ITfr in general and Anabasis in particular are threatened by climate change at least in the parts of the ITfr that will experience higher precipitation in the future, and because of that the conservation of those ITfr habitats needs to be prioritized.
In summary, an extensive sampling of Anabasis (21 out of 28 species included in the molecular analyses) revealed the complex biogeography of the genus and showed that species occurring in the same floristic region do not form monophyletic groups but are a mosaic of old and young lineages of this genus. Like other xerophytic elements of the ITfr, Anabasis diversified during the late Miocene spread into the adjacent arid biomes of Asia and North Africa. As has been shown for Haplophyllum, the ITfr was identified as cradle for some arid-adapted taxa of Asia and North Africa, if it is also a sink area for the arid-adapted lineage Anabasis remains ambiguous. The proposed hypothesis that the expansion of Anabasis coincides with the spread of arid and semi-arid biomes in Eurasia needs to be rejected.
Anabasis did not follow aridification and continuously expanded its distribution area, in fact its ancestral distribution area seems to have been fragmented during the very Late Miocene and the remnant lineages then expanded into neighboring arid regions.

ACK N OWLED G M ENTS
We thank Rüdiger Masson for generating part of the sequence data and his careful taxonomic study of the Anabasis specimens.
We thank Helmut Freitag for material and helpful advice at the be-

DATA ACCE SS I B I LIT Y
Genbank accessions MF156717-MF156846 and MF580497-

MF580548 (for further information see Supporting Information
Appendix S1).